![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Errant thought from June 13:
The world keeps seeming more and more dystopian. But it is really amazing to be able to ask a computer complex things in natural language, and get back coherent, valid responses in natural language too. My mind often glosses over that wonder because of everything else going on. And also because of knowing that wonderful achievement has its own associated dystopian aspects to it, which will just get worse and worse as they are exploited.
.
Like this: The AI Slop Fight Between Iran and Israel
.
Kids growing up now and in the future may find it difficult to believe actual photos and movie footage from past atrocities, including the Holocaust. They will have no way to trust that the images are real and were not generated by AI. Perhaps if they looked through an old book with photos, and could trust that it was published in the year it said it was, before AI was capable of what it is now, perhaps that would be believable to them. But books are becoming rarer, especially old ones. Old books didn't have many photos. Were there many books with photos of atrocities? Encyclopedias would have perhaps a few photos for each topic. Certain magazines were more likely to have extensive photos. There may be digitized versions of the books and magazines available online, but then you get back to how can you trust that the digitized item you're looking at isn't a fake?
I'm already at that distrusting stage with much of what I see online. When you don't know what is trustworthy, you end up choosing to believe the things which align with your already-formed beliefs. "This aligns with my beliefs; I don't know if it is true, but I wouldn't be surprised if it was." "This doesn't align with my beliefs; why spend time and mental effort considering that it might be true when it very well might not be?"
The world keeps seeming more and more dystopian. But it is really amazing to be able to ask a computer complex things in natural language, and get back coherent, valid responses in natural language too. My mind often glosses over that wonder because of everything else going on. And also because of knowing that wonderful achievement has its own associated dystopian aspects to it, which will just get worse and worse as they are exploited.
.
Like this: The AI Slop Fight Between Iran and Israel
.
Kids growing up now and in the future may find it difficult to believe actual photos and movie footage from past atrocities, including the Holocaust. They will have no way to trust that the images are real and were not generated by AI. Perhaps if they looked through an old book with photos, and could trust that it was published in the year it said it was, before AI was capable of what it is now, perhaps that would be believable to them. But books are becoming rarer, especially old ones. Old books didn't have many photos. Were there many books with photos of atrocities? Encyclopedias would have perhaps a few photos for each topic. Certain magazines were more likely to have extensive photos. There may be digitized versions of the books and magazines available online, but then you get back to how can you trust that the digitized item you're looking at isn't a fake?
I'm already at that distrusting stage with much of what I see online. When you don't know what is trustworthy, you end up choosing to believe the things which align with your already-formed beliefs. "This aligns with my beliefs; I don't know if it is true, but I wouldn't be surprised if it was." "This doesn't align with my beliefs; why spend time and mental effort considering that it might be true when it very well might not be?"
no subject
Date: 2025-06-28 11:22 pm (UTC)From:I imagine a cottage industry of people checking to see which pre-AI publications are faked by AI.
Fake media has been in existence for a long time. It has sometimes taken scholars ages to identify that a "medieval manuscript" actually isn't. And certain regimes have been at pains to rewrite the history of their countries. But the kind of world we're entering into, in which the entire past is being rewritten . . .
I don't know the answer to this crisis, unfortunately. For now, I do think that pre-AI media serves an important role. So does knowing who has been trustworthy in the past. But these are scary times.
no subject
Date: 2025-06-30 07:03 am (UTC)From:But another thing I was thinking of today, was about how I've been using AI chat more and more to get answers to things, when I can't easily find them the other ways. I've used MS Copilot the most so far because its answers usually seem good, and it doesn't make me login. But I was thinking, if people in the future get all their answers from their favorite LLM for everything, it's similar to watching only one news channel (ie Fox News or CNN, etc.). I'm sure the Fox News people either have such an LLM already, or will get one, trained on data from its sites and forums. It probably won't refuse to talk about politics like some of the others.
no subject
Date: 2025-06-30 09:49 am (UTC)From:Thanks for the PM!
"And 'rewriting' history can make sense if the old information was biased and/or new info comes to light even if it wasn't fake per se originally."
What I had in mind was regimes deliberately faking the past, like erasing someone from a photo because he has fallen out of favor with the Powers To Be. But certainly the line between faking history and correcting past historians can be a thin one, particularly when it takes the form of omitting information.
LLMs: I'm afraid I don't trust their reliability yet. [Deletes long rant on them, since you probably know more on this subject than I do.]